Reconstruction of movies of facial expressions

Abstract

We present a technique for reconstructing facial movies, given a small number of real images and a few parameters for the in-between images. The parameters can be automatically extracted by a tracking system. This scheme can also be used for creating realistic facial animations and for compression of facial movies. The in-between images are produced without ever generating a 3D model of the face. Since facial motions due to expressions are not well defined mathematically our approach is based on utilizing image patterns in facial optical flow. These patterns were revealed by an empirical study which analyzed and compared image motion patterns in facial expressions.